To compare two different routes of intermittently administered rhIL-2 with a highly active antiretroviral regimen to highly active antiretroviral therapy alone with respect to the proportion of subjects achieving 50% increase in CD4 counts above prerandomization baseline values after one year of rhIL-2 and the rate of change in CD4 counts. To compare the safety tolerance and effect on quality of life of these regimens. To assess the effects of rhIL-2 with highly active antiretroviral therapy on changes in immune cell phenotypes and function and on HIV viral load and rate of antiviral drung resistance development.